GDBN: a Graph Neural Network Approach to Dynamic Bayesian Network

TL;DR

This paper introduces GDBN, a graph neural network method for learning sparse causal structures in dynamic systems, outperforming existing approaches in accuracy and inference quality.

Contribution

It proposes a novel GNN-based score method for causal discovery in dynamic Bayesian networks, improving accuracy over linear models and state-of-the-art techniques.

Findings

GDBN significantly outperforms other methods in causal inference accuracy.

The structural causal model learned by GDBN is more precise than linear SCMs.

Experimental results validate the effectiveness of the proposed approach.

Abstract

Identifying causal relations among multi-variate time series is one of the most important elements towards understanding the complex mechanisms underlying the dynamic system. It provides critical tools for forecasting, simulations and interventions in science and business analytics. In this paper, we proposed a graph neural network approach with score-based method aiming at learning a sparse DAG that captures the causal dependencies in a discretized time temporal graph. We demonstrate methods with graph neural network significantly outperformed other state-of-the-art methods with dynamic bayesian networking inference. In addition, from the experiments, the structural causal model can be more accurate than a linear SCM discovered by the methods such as Notears.